Dataset Provided by : Dhruvil Dave - https://www.kaggle.com/dhruvildave/top-play-store-games¶

Import Libraries¶

In [ ]:
import pandas as pd
import numpy as np
import os

import matplotlib.pyplot as plt
import plotly.io as pio
import plotly.express as px
import seaborn as sns
import nbformat
In [ ]:
pwd = os.getcwd() 
filepath = pwd + "/android-games.csv"
df = pd.read_csv(filepath)

Viewing the Dataset¶

lets read and check our dataset

In [ ]:
print ('Data shape is: ', df.shape)

Data shape is:  (1730, 15)
  • so we have 1730 rows and 15 columns working with
  • lets take a look at the columns and general summary info.
In [ ]:
pd.DataFrame({"Columns": df.columns})
Out[ ]:
Columns
0 rank
1 title
2 total ratings
3 installs
4 average rating
5 growth (30 days)
6 growth (60 days)
7 price
8 category
9 5 star ratings
10 4 star ratings
11 3 star ratings
12 2 star ratings
13 1 star ratings
14 paid
In [ ]:
df.isnull().sum()
Out[ ]:
rank                0
title               0
total ratings       0
installs            0
average rating      0
growth (30 days)    0
growth (60 days)    0
price               0
category            0
5 star ratings      0
4 star ratings      0
3 star ratings      0
2 star ratings      0
1 star ratings      0
paid                0
dtype: int64

Great there are no missing vlaues within this dataset. That saves us some time from dealing with NAN's.

  • Lets sample our dataset to explore what it contains.
  • lets sample the first 8 rows.
In [ ]:
df.sample(n=8)
Out[ ]:
rank title total ratings installs average rating growth (30 days) growth (60 days) price category 5 star ratings 4 star ratings 3 star ratings 2 star ratings 1 star ratings paid
317 18 Checkers 554532 100.0 M 4 1.4 3.0 0.0 GAME BOARD 371827 69515 37892 18726 56569 False
601 76 ไพ่แคง-รวมดัมมี่dummy ป๊อกเด้ง เก้าเกไทย เกมไพ... 120434 1.0 M 4 1.5 3.0 0.0 GAME CASINO 104694 7396 4236 1375 2731 False
1331 6 Carrom Pool: Disc Game 3181731 100.0 M 4 7.4 11.5 0.0 GAME SPORTS 2157840 364339 220366 106843 332341 False
1022 97 Cut the Rope FULL FREE 390519 100.0 M 4 0.0 0.1 0.0 GAME PUZZLE 304567 48911 16267 5635 15137 False
985 60 Easy Game - Brain Test and Tricky Mind Puzzles 547548 10.0 M 4 1.1 2.5 0.0 GAME PUZZLE 389864 72820 30413 13849 40600 False
666 41 Save The Girl 952348 100.0 M 3 1.1 2.6 0.0 GAME CASUAL 591803 80978 62429 48097 169038 False
1174 49 School of Chaos Online MMORPG 503643 10.0 M 4 0.2 0.5 0.0 GAME ROLE PLAYING 335781 35856 25781 18113 88109 False
1094 69 Boomerang Make and Race - Scooby-Doo Racing Game 380849 50.0 M 4 0.4 1.0 0.0 GAME RACING 279496 26949 18355 13010 43037 False
In [ ]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1730 entries, 0 to 1729
Data columns (total 15 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   rank              1730 non-null   int64  
 1   title             1730 non-null   object 
 2   total ratings     1730 non-null   int64  
 3   installs          1730 non-null   object 
 4   average rating    1730 non-null   int64  
 5   growth (30 days)  1730 non-null   float64
 6   growth (60 days)  1730 non-null   float64
 7   price             1730 non-null   float64
 8   category          1730 non-null   object 
 9   5 star ratings    1730 non-null   int64  
 10  4 star ratings    1730 non-null   int64  
 11  3 star ratings    1730 non-null   int64  
 12  2 star ratings    1730 non-null   int64  
 13  1 star ratings    1730 non-null   int64  
 14  paid              1730 non-null   bool   
dtypes: bool(1), float64(3), int64(8), object(3)
memory usage: 191.0+ KB

The install column can be useful for analysis. However note that it's a object type . It may be more useful if it were an integer (int) or float.

  • May need to reformat this later on.

Summary so far:¶

  • The dataset consists of games from different categories, ratings and number of installs.
  • there are no NAN / null values.
  • installs can be useful for analysis. It would be good to reformat it to a numerical variable.
  • category is a categorical variable, making it useful for groupings.
  • numerical varibles should be paid close attentions to.
  • should some columns be dropped ?
In [ ]:
df['price'].value_counts()
Out[ ]:
0.00    1723
1.99       3
7.49       1
2.99       1
1.49       1
0.99       1
Name: price, dtype: int64
In [ ]:
df['paid'].value_counts()
Out[ ]:
False    1723
True        7
Name: paid, dtype: int64

Dropping Price column¶

we can drop the price column since most games are free and only 7 aren't, also the price of a game isn't relevant to the EDA.

In [ ]:
df.drop('price', axis=1, inplace=True)
In [ ]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1730 entries, 0 to 1729
Data columns (total 14 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   rank              1730 non-null   int64  
 1   title             1730 non-null   object 
 2   total ratings     1730 non-null   int64  
 3   installs          1730 non-null   object 
 4   average rating    1730 non-null   int64  
 5   growth (30 days)  1730 non-null   float64
 6   growth (60 days)  1730 non-null   float64
 7   category          1730 non-null   object 
 8   5 star ratings    1730 non-null   int64  
 9   4 star ratings    1730 non-null   int64  
 10  3 star ratings    1730 non-null   int64  
 11  2 star ratings    1730 non-null   int64  
 12  1 star ratings    1730 non-null   int64  
 13  paid              1730 non-null   bool   
dtypes: bool(1), float64(2), int64(8), object(3)
memory usage: 177.5+ KB

Categories Column¶

In [ ]:
df['category'].value_counts()
Out[ ]:
GAME CARD            126
GAME WORD            104
GAME PUZZLE          100
GAME TRIVIA          100
GAME STRATEGY        100
GAME SPORTS          100
GAME SIMULATION      100
GAME ROLE PLAYING    100
GAME RACING          100
GAME ACTION          100
GAME ADVENTURE       100
GAME EDUCATIONAL     100
GAME CASUAL          100
GAME CASINO          100
GAME BOARD           100
GAME ARCADE          100
GAME MUSIC           100
Name: category, dtype: int64
In [ ]:
df['category'].value_counts(ascending=True).plot(kind='barh', title='Game Categories')
Out[ ]:
<AxesSubplot:title={'center':'Game Categories'}>
In [ ]:
plt.figure(figsize=(25,5))
sns.set_theme(style="darkgrid")
sns.countplot(df['category'])
plt.title('Number of Games by Category')
plt.xlabel('game category')
plt.ylabel('# of games')

/Users/rojaywhite/opt/miniconda3/envs/UdemyPython/lib/python3.9/site-packages/seaborn/_decorators.py:36: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.
  warnings.warn(
Out[ ]:
Text(0, 0.5, '# of games')
In [ ]:
# using the plotly.express library
pio.renderers.default = "notebook" #sets the plotly default render.

fig = px.histogram(df, x="category", title="Game Categories", labels={"category": "Categories"})
fig.update_layout(xaxis={"categoryorder":"total descending"},)
fig.update_yaxes(automargin=True)
fig.show()

Paid vs Free games¶

In [ ]:
df['paid'].value_counts()
Out[ ]:
False    1723
True        7
Name: paid, dtype: int64
In [ ]:
df['paid'].value_counts().plot(kind='pie', title='Free vs Paid games', legend=True,
    autopct='%1.1f%%', fontsize=14, figsize=[7, 7], labels=["free", "paid"], ylabel='', colors=["g", "r"])
Out[ ]:
<AxesSubplot:title={'center':'Free vs Paid games'}>
In [ ]:
free_games = df['paid'].value_counts()
label = ['Free', 'Paid']
values = df['paid'].value_counts().values

fig = px.pie(free_games, values=values, names=label, title='Free vs Paid games', color_discrete_sequence=["springgreen", "red"])
fig.update_traces(textposition='outside', textinfo='percent+label')
fig.show()

Relationship Analysis¶

In [ ]:
df.head(n=5)
Out[ ]:
rank title total ratings installs average rating growth (30 days) growth (60 days) category 5 star ratings 4 star ratings 3 star ratings 2 star ratings 1 star ratings paid
0 1 Garena Free Fire- World Series 86273129 500.0 M 4 2.1 6.9 GAME ACTION 63546766 4949507 3158756 2122183 12495915 False
1 2 PUBG MOBILE - Traverse 37276732 500.0 M 4 1.8 3.6 GAME ACTION 28339753 2164478 1253185 809821 4709492 False
2 3 Mobile Legends: Bang Bang 26663595 100.0 M 4 1.5 3.2 GAME ACTION 18777988 1812094 1050600 713912 4308998 False
3 4 Brawl Stars 17971552 100.0 M 4 1.4 4.4 GAME ACTION 13018610 1552950 774012 406184 2219794 False
4 5 Sniper 3D: Fun Free Online FPS Shooting Game 14464235 500.0 M 4 0.8 1.5 GAME ACTION 9827328 2124154 1047741 380670 1084340 False
In [ ]:
df.describe()
Out[ ]:
rank total ratings average rating growth (30 days) growth (60 days) 5 star ratings 4 star ratings 3 star ratings 2 star ratings 1 star ratings
count 1730.000000 1.730000e+03 1730.000000 1730.000000 1730.000000 1.730000e+03 1.730000e+03 1.730000e+03 1.730000e+03 1.730000e+03
mean 50.386705 1.064332e+06 3.908092 321.735896 122.554971 7.622315e+05 1.164366e+05 5.706307e+04 2.710336e+04 1.014950e+05
std 28.936742 3.429250e+06 0.290973 6018.914507 2253.891703 2.538658e+06 3.021631e+05 1.495314e+05 8.154542e+04 4.083745e+05
min 1.000000 3.299300e+04 2.000000 0.000000 0.000000 1.397500e+04 2.451000e+03 7.180000e+02 2.660000e+02 5.450000e+02
25% 25.000000 1.759992e+05 4.000000 0.100000 0.200000 1.277300e+05 2.064300e+04 9.652500e+03 4.262250e+03 1.281200e+04
50% 50.000000 4.286065e+05 4.000000 0.500000 1.000000 2.964340e+05 5.098050e+04 2.507800e+04 1.067550e+04 3.368600e+04
75% 75.000000 8.837970e+05 4.000000 1.700000 3.300000 6.198358e+05 1.018140e+05 5.229500e+04 2.322875e+04 8.015725e+04
max 100.000000 8.627313e+07 4.000000 227105.700000 69441.400000 6.354677e+07 5.404966e+06 3.158756e+06 2.122183e+06 1.249592e+07

Grouping¶

In [ ]:
total_ratings_by_cateogry = df.groupby('category')['total ratings'].mean()
total_ratings_by_cateogry
Out[ ]:
category
GAME ACTION          4.011344e+06
GAME ADVENTURE       8.935617e+05
GAME ARCADE          1.793780e+06
GAME BOARD           4.457431e+05
GAME CARD            3.326041e+05
GAME CASINO          3.619031e+05
GAME CASUAL          2.470866e+06
GAME EDUCATIONAL     1.529804e+05
GAME MUSIC           2.163020e+05
GAME PUZZLE          9.466929e+05
GAME RACING          1.139027e+06
GAME ROLE PLAYING    7.087648e+05
GAME SIMULATION      9.341417e+05
GAME SPORTS          1.353829e+06
GAME STRATEGY        1.856570e+06
GAME TRIVIA          2.982217e+05
GAME WORD            3.943603e+05
Name: total ratings, dtype: float64

N.B. Think of groupby() splits the dataset data into buckets by 'category', and then splitting the records inside each category bucket by the 'total ratings'.

Total Ratings grouped by Category¶

In [ ]:
fig = px.bar(total_ratings_by_cateogry, x=total_ratings_by_cateogry.index, y=total_ratings_by_cateogry.values,
             labels={'y': 'Ratings'})
fig.update_layout(xaxis={'categoryorder':'total descending'})
fig.show()
In [ ]:
df['installs']
Out[ ]:
0       500.0 M
1       500.0 M
2       100.0 M
3       100.0 M
4       500.0 M
         ...   
1725      1.0 M
1726      1.0 M
1727     10.0 M
1728      5.0 M
1729     10.0 M
Name: installs, Length: 1730, dtype: object

need to convert this column to a numerical type.

In [ ]:
def numbers(df):
    if df.split(".")[1].split(" ")[1] == "M":
        return int(df.split(".")[0])
    else:
        return int(df.split(".")[0])/1000
df["installs"] = df.installs.apply(numbers)
df = df.rename(columns={'installs': 'installs_in_million'})
In [ ]:
df["installs_in_million"].head()

it's now a type float.

Installs in Millions grouped by Category¶

In [ ]:
intalls_in_million_by_category = df.groupby('category')['installs_in_million'].mean()
intalls_in_million_by_category
In [ ]:
fig = px.bar(intalls_in_million_by_category, x=intalls_in_million_by_category.index, y=intalls_in_million_by_category.values,
            width=700, height=400,labels={"y":"Total Installs in Million"})
fig.update_layout(xaxis={"categoryorder":"total descending"}),
margin=dict(l=10, r=10, t=10, b=10)
fig.show()

Growth (30 days) grouped by Category¶

In [ ]:
df.sample(n=4)
In [ ]:
growth = df.groupby('category')[['growth (30 days)',"growth (60 days)"]].mean()
growth
In [ ]:
fig = px.bar(growth, y="growth (30 days)",
labels={"y":"Growth (30 Days)", "category": "Category", "value":"Total Growth"})
fig.update_layout(xaxis={"categoryorder":"total descending"})
In [ ]:
fig = px.bar(growth,y="growth (60 days)",
labels={"y":"Growth (60 Days)", "category":"Category", "value":"Total Growth"})
fig.update_layout(xaxis={"categoryorder":"total descending"})
In [ ]:
fig = px.line(growth,  y=["growth (30 days)", "growth (60 days)"],title="Games Growth",
labels={"category":"Category", "values":"Total Growth"})
fig.show()